Token导航 LogoToken导航TokenDH.com
研究检索需要联网clawhub未标认证来源可访问clear审计通过

self-healing-agent自愈剂

Agent Skill

self-healing-agent 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

7,962

周安装

342

GitHub Stars

公开资料未说明

下载量

2,791
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:self-healing-agent(自愈剂)
来源仓库:https://github.com/stevojarvisai-star/self-healing-agent
安装命令:
openclaw skills install self-healing-agent
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install self-healing-agent

简介

监控代理健康状况并自动修复崩溃作业、损坏技能等问题。

  • 适用于长时间运行Agent的运维支持与异常自恢复场景。
  • 覆盖cron作业、内存泄漏、配置损坏等多类故障类型。
  • 修复操作前应验证备份完整性,避免二次损坏。
  • self-healing-agent 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

name
self-healing-agent
description
Self-recovery and auto-repair system for OpenClaw agents. Monitors agent health, detects failures (crashed cron jobs, broken skills, config corruption, memory bloat, session timeouts), automatically diagnoses root causes, and applies fixes. Inspired by Hermes agent's self-recovery capabilities. Use when asked to "make my agent self-healing", "auto-repair", "self-recovery", "agent health monitoring", "fix broken cron jobs automatically", "auto-diagnose failures", "resilient agent", "fault tolerance", "watchdog", "heartbeat monitor", or when agents repeatedly fail and need automated recovery. Also use to set up continuous health monitoring via cron.

Self-Healing Agent

Automated failure detection, diagnosis, and recovery for OpenClaw agents. The watchdog that keeps your agent running.

Quick Start

# Full health check — scan all systems, diagnose issues, suggest fixes
python3 scripts/self-healing-agent.py check

# Auto-heal — detect and fix what it can automatically
python3 scripts/self-healing-agent.py heal

# Monitor mode — run continuously, fix issues as they appear
python3 scripts/self-healing-agent.py monitor --interval 300

# Check specific subsystem
python3 scripts/self-healing-agent.py check --target cron
python3 scripts/self-healing-agent.py check --target memory
python3 scripts/self-healing-agent.py check --target config
python3 scripts/self-healing-agent.py check --target sessions

Commands

check — Health Check

Runs diagnostic suite:

  • Cron health — Failed jobs, consecutive errors, stuck jobs, timeout patterns
  • Memory health — File sizes, bloated sessions, orphaned files, growth rate
  • Config health — Valid JSON, required fields present, deprecated settings
  • Session health — Active sessions, zombie processes, memory usage
  • Skill health — Broken scripts, missing dependencies, syntax errors
  • Network health — API endpoint reachability, DNS resolution, SSL cert validity

Options: --target <subsystem> to check one area, --json for machine output.

heal — Auto-Repair

For each detected issue, applies the safest fix:

  • Restarts failed cron jobs (after diagnosing root cause)
  • Clears bloated session files (with backup)
  • Fixes JSON syntax errors in config (common trailing comma, etc.)
  • Removes orphaned process files
  • Restores corrupted memory files from git history
  • Rotates oversized log files

Options: --dry-run to preview, --aggressive for riskier fixes.

monitor — Continuous Watchdog

Runs in a loop, checking health every N seconds:

  • Logs findings to memory/self-healing-log.json
  • Auto-heals fixable issues
  • Escalates unfixable issues to the agent's main session
  • Tracks MTTR (mean time to recovery) and failure patterns

Options: --interval <seconds> (default: 300), --max-heals <n> per cycle.

report — Health Report

Generates a markdown health report covering:

  • Last 24h failure count and types
  • MTTR statistics
  • Most common failure modes
  • Recommendations for prevention

What It Monitors

SubsystemChecksAuto-Heals
CronFailed runs, timeouts, consecutive errorsRestart jobs, clear error state
MemoryFile sizes >1MB, growth rate, duplicatesArchive old files, compact
ConfigJSON validity, required fields, deprecated keysFix syntax, add defaults
SessionsZombie processes, bloated contextsKill zombies, archive contexts
SkillsSyntax errors, missing deps, broken importsLog issue, skip broken skill
NetworkAPI endpoints, DNS, SSL certsRetry with backoff, switch endpoints

Healing Log

All actions are logged to memory/self-healing-log.json:

{
  "timestamp": "2026-04-05T06:00:00Z",
  "issue": "cron job 'daily-intel' failed 3 consecutive times",
  "diagnosis": "Script timeout — API rate limit hit",
  "action": "Reset error count, added 30s backoff, restarted",
  "result": "success",
  "mttr_seconds": 12
}

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

71.35%
按下载量换算1,991

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills